Improving priors for USTC via transmission tomography: A simulation study

Abstract

The current USCT-algorithms have limitations due to the object size and contrast to be reconstructed. To avoid USCT-algorithms divergence, it is common to initialize USCT-algorithms with a priori anatomical information of the reconstructing region. This information can be obtained from transmitted signals. Our work presents, on simulated medium, an alternative to the Modified Median Filter (MMF) that improves the transmission reconstructed image, and aims to offer a better a priori information to USCT-algorithms. Herein, it was used: 64 transducers (100 kHz) distributed around the medium; Two sets of data were generated: the first with speed of sound between [1400 1680]m/s and the second with speed of sound between [1350 1730]m/s simulating a higher contrast medium; Simultaneous Algebraic Reconstruction Technique (SIRT) as the transmission reconstruction algorithm; a Matlab toolbox (k-wave) for data generation of ultrasound propagation on heterogeneous medium; Spatial filters (mean, median, Gaussian), the original MMF over several thresholds and its modification proposed here (m-MMF), which eliminates the need of threshold. Normalized Root Mean Square Error was utilized over the reconstructing objects for evaluation. Comparing the SIRT reconstructions performance, better results were found when any studied filter is executed with the SIRT compared to none. Among the filters, the m-MMF yielded better results than the other studied filters, and more importantly, the improvements were higher for high contrast. The investigation suggests that the use of specific filters on Transmission reconstruction algorithm for USCT improve the extraction of information of the medium that might be valuable as priors for USCT of high contrast heterogeneous medium.